Posts in category .net - page 5

Multiple outputs from T4 made easy – revisited

My multiple outputs from t4 made easy post contained a class making it easy to produce multiple files from Visual Studio’s text templating engine (T4).

While useful it had a few issues:

  • Getting start/end blocks mixed up resulted in unpredictable behavior
  • Files were rewritten even when content did not change
  • Did not play well with source control
  • Files not always deleted in VS
  • Failed in Visual Studio’s project-less Web Sites

This helper class forms the basis of multiple file output for Entity Framework templates in .NET 4.0 and the LINQ to SQL templates on CodePlex so we (Jeff Reed, Andrew Peters and myself) made the following changes.

Improvements

Simpler block handling

The header, footer and file blocks can now be completed with EndBlock (EndHeader and EndFooter are gone), although it will automatically end the previous block when it hits a new one or the final Process method.

Skip unchanged files

Files are now only written to disk if the contents are different with the exception of the original T4 output file (we can’t stop that, sorry).

There is additional overhead reading and comparing files we believe unmodified files keeping their dates and source control status are worth it.

Automatic checkout

When the template detects it is running in Visual Studio and that the file it needs to write to is currently in source control but not checked out it will check the file out for you.

Predictable clean-up

All files that were not part of the generation process but are nested under the project item will now be deleted when running inside Visual Studio.

Outside of Visual Studio files are no longer deleted – this was destructive and it couldn’t know which files it generated on a previous run to clean-up correctly anyway.

Website projects fall back to single file generation

Visual Studio has both web sites and web applications with the former being project-less leading to very messy multi-file generation so it forces single file generation.

Internal improvements

Source is now simpler to read and understand with less public visibility and faster and more robust VS interop by batching the files & deletes to a single invoke at the end to avoid conflicts with other add-ins that might be triggered by the changes.

Usage

Initialization

You’ll need to get the code into your template – either copy the code in or reference it with an include directive. Then declare an instance of the Manager class passing in some environmental options such as the desired default output path. (For Visual Studio 2010 remove the #v3.5 portion from the language attribute)

<#@ template language="C#v3.5" hostspecific="True"
#><#@include file="Manager.ttinclude"
#><# var manager = Manager.Create(Host, GenerationEnvironment); #>

File blocks

Then add one line before and one line after each block which could be split out into it’s own file passing in what the filename would be if split. The EndBlock is optional if you want it to carry through to the next one :)

<# manager.StartNewFile("Employee.generated.cs"); #>
public class Employee {  }
<# manager.EndBlock(); #>

Headers & footers

Many templates need to share a common header/footer for such things as comments or using/import statements or turning on/off warnings. Simply use StartHeader and StartFooter and the blocks will be emitted to the start and end of all split files as well as being left in the original output file.

<# manager.StartHeader(); #>
// Code generated by a template
using System;

<# manager.EndBlock(); #>

Process

At the end of the template call Process to handle splitting the files (true) or not (false). Anything not included in a specific StartNewFile block will remain in the original output file.

<# manager.Process(true); #>

Revised Manager class

Latest source available at GitHub
<#@ assembly name="System.Core"
#><#@ assembly name="System.Data.Linq"
#><#@ assembly name="EnvDTE"
#><#@ assembly name="System.Xml"
#><#@ assembly name="System.Xml.Linq"
#><#@ import namespace="System"
#><#@ import namespace="System.CodeDom"
#><#@ import namespace="System.CodeDom.Compiler"
#><#@ import namespace="System.Collections.Generic"
#><#@ import namespace="System.Data.Linq"
#><#@ import namespace="System.Data.Linq.Mapping"
#><#@ import namespace="System.IO"
#><#@ import namespace="System.Linq"
#><#@ import namespace="System.Reflection"
#><#@ import namespace="System.Text"
#><#@ import namespace="System.Xml.Linq"
#><#@ import namespace="Microsoft.VisualStudio.TextTemplating"
#><#+

// Manager class records the various blocks so it can split them up
class Manager {
    private class Block {
        public String Name;
        public int Start, Length;
    }

    private Block currentBlock;
    private List<Block> files = new List<Block>();
    private Block footer = new Block();
    private Block header = new Block();
    private ITextTemplatingEngineHost host;
    private StringBuilder template;
    protected List<String> generatedFileNames = new List<String>();

    public static Manager Create(ITextTemplatingEngineHost host, StringBuilder template) {
        return (host is IServiceProvider) ? new VSManager(host, template) : new Manager(host, template);
    }

    public void StartNewFile(String name) {
        if (name == null)
            throw new ArgumentNullException("name");
        CurrentBlock = new Block { Name = name };
    }

    public void StartFooter() {
        CurrentBlock = footer;
    }

    public void StartHeader() {
        CurrentBlock = header;
    }

    public void EndBlock() {
        if (CurrentBlock == null)
            return;
        CurrentBlock.Length = template.Length - CurrentBlock.Start;
        if (CurrentBlock != header && CurrentBlock != footer)
            files.Add(CurrentBlock);
        currentBlock = null;
    }

    public virtual void Process(bool split) {
        if (split) {
            EndBlock();
            String headerText = template.ToString(header.Start, header.Length);
            String footerText = template.ToString(footer.Start, footer.Length);
            String outputPath = Path.GetDirectoryName(host.TemplateFile);
            files.Reverse();
            foreach(Block block in files) {
                String fileName = Path.Combine(outputPath, block.Name);
                String content = headerText + template.ToString(block.Start, block.Length) + footerText;
                generatedFileNames.Add(fileName);
                CreateFile(fileName, content);
                template.Remove(block.Start, block.Length);
            }
        }
    }

    protected virtual void CreateFile(String fileName, String content) {
        if (IsFileContentDifferent(fileName, content))
            File.WriteAllText(fileName, content);
    }

    public virtual String GetCustomToolNamespace(String fileName) {
        return null;
    }

    public virtual String DefaultProjectNamespace {
        get { return null; }
    }

    protected bool IsFileContentDifferent(String fileName, String newContent) {
        return !(File.Exists(fileName) && File.ReadAllText(fileName) == newContent);
    }

    private Manager(ITextTemplatingEngineHost host, StringBuilder template) {
        this.host = host;
        this.template = template;
    }

    private Block CurrentBlock {
        get { return currentBlock; }
        set {
            if (CurrentBlock != null)
                EndBlock();
            if (value != null)
                value.Start = template.Length;
            currentBlock = value;
        }
    }

    private class VSManager: Manager {
        private EnvDTE.ProjectItem templateProjectItem;
        private EnvDTE.DTE dte;
        private Action<String> checkOutAction;
        private Action<IEnumerable<String>> projectSyncAction;

        public override String DefaultProjectNamespace {
            get {
                return templateProjectItem.ContainingProject.Properties.Item("DefaultNamespace").Value.ToString();
            }
        }

        public override String GetCustomToolNamespace(string fileName) {
            return dte.Solution.FindProjectItem(fileName).Properties.Item("CustomToolNamespace").Value.ToString();
        }

        public override void Process(bool split) {
            if (templateProjectItem.ProjectItems == null)
                return;
            base.Process(split);
            projectSyncAction.EndInvoke(projectSyncAction.BeginInvoke(generatedFileNames, null, null));
        }

        protected override void CreateFile(String fileName, String content) {
            if (IsFileContentDifferent(fileName, content)) {
                CheckoutFileIfRequired(fileName);
                File.WriteAllText(fileName, content);
            }
        }

        internal VSManager(ITextTemplatingEngineHost host, StringBuilder template)
            : base(host, template) {
            var hostServiceProvider = (IServiceProvider) host;
            if (hostServiceProvider == null)
                throw new ArgumentNullException("Could not obtain IServiceProvider");
            dte = (EnvDTE.DTE) hostServiceProvider.GetService(typeof(EnvDTE.DTE));
            if (dte == null)
                throw new ArgumentNullException("Could not obtain DTE from host");
            templateProjectItem = dte.Solution.FindProjectItem(host.TemplateFile);
            checkOutAction = (String fileName) => dte.SourceControl.CheckOutItem(fileName);
            projectSyncAction = (IEnumerable<String> keepFileNames) => ProjectSync(templateProjectItem, keepFileNames);
        }

        private static void ProjectSync(EnvDTE.ProjectItem templateProjectItem, IEnumerable<String> keepFileNames) {
            var keepFileNameSet = new HashSet<String>(keepFileNames);
            var projectFiles = new Dictionary<String, EnvDTE.ProjectItem>();
            var originalFilePrefix = Path.GetFileNameWithoutExtension(templateProjectItem.get_FileNames(0)) + ".";
            foreach(EnvDTE.ProjectItem projectItem in templateProjectItem.ProjectItems)
                projectFiles.Add(projectItem.get_FileNames(0), projectItem);

            // Remove unused items from the project
            foreach(var pair in projectFiles)
                if (!keepFileNames.Contains(pair.Key) && !(Path.GetFileNameWithoutExtension(pair.Key) + ".").StartsWith(originalFilePrefix))
                    pair.Value.Delete();

            // Add missing files to the project
            foreach(String fileName in keepFileNameSet)
                if (!projectFiles.ContainsKey(fileName))
                    templateProjectItem.ProjectItems.AddFromFile(fileName);
        }

        private void CheckoutFileIfRequired(String fileName) {
            var sc = dte.SourceControl;
            if (sc != null && sc.IsItemUnderSCC(fileName) && !sc.IsItemCheckedOut(fileName))
                checkOutAction.EndInvoke(checkOutAction.BeginInvoke(fileName, null, null));
        }
    }
} #>

[)amien

When an object-relational mapper is too much, DataReader too little

I fired up Visual Studio this evening to write a proof-of-concept app and found myself wanting strongly typed domain objects from a database but without the overhead of an object-relational mapper  (the application is read-only).

One solution is to write methods by hand, another is to code generate them but it would be nice to be able to do:

var customers = new SqlCommand("SELECT ID, Name FROM Customer", connection)
  .As(r => new Customer { CustomerID = r.GetInt32(0), Name = r.GetString(1) }).ToList();

So for any DbCommand object you can turn it into a bunch of classes by specifying the new pattern.

The tiny helper class to achieve this is:

public static class DataHelpers {
  public static List<T> ToList<T>(this IEnumerable<T> enumerable) {
    return new List<T>(enumerable);
  }

  public static IEnumerable<T> As<T>(this DbCommand command, Func<IDataRecord, T> map) {
    using (var reader = command.ExecuteReader())
      while (reader.Read())
        yield return map(reader);
  }
}

It might even be possible to do some cool caching/materialization. I should look into that :)

[)amien

LINQ to SQL cheat sheet

Thumbnail of the LINQ to SQL Cheat Sheet PDF

A few short words to say I’ve put together a cheat sheet for LINQ to SQL with one page for C# and another for VB.NET.

It shows the syntax for a number of common query operations, manipulations and attributes and can be a very useful quick reference :)

Download LINQ to SQL cheat sheet (PDF) (76 KB)

[)amien

Dictionary look-up or create made simpler

The design of a Dictionary lends itself well to a caching or identification mechanism and as a result you often see code that looks like this:

private static Dictionary<string, Employee> employees = new Dictionary<string, Employee>();

public static Employee GetByName(string name) {
  Employee employee;
  if (!employees.TryGetValue(name, out employee)) {
    employee = new Employee(whatever);
    employees.Add(name, employee);
  }
  return employee;
}

It’s not that it is particularly difficult but it can be a bit error prone and when you’re doing it over and over. What would be nicer is something that let you do:

public static Employee GetByName(string name) {
  return employees.GetOrAdd(name, () => new Employee(whatever));
}

Here’s an extension method to drop-in to a static class of your choosing that achieves just that.

public static TDictionaryValue GetOrAdd<TKey, TDictionaryValue>(
  this IDictionary<TKey, TDictionaryValue> dictionary,
  TKey key,
  Func<TDictionaryValue> newValue
) {
  TDictionaryValue value;
  if (!dictionary.TryGetValue(key, out value)) {
    value = newValue.Invoke();
    dictionary.Add(key, value);
  }
  return value;
}

[)amien